• 한국재난정보학회 논문집
• /
• 제11권3호
• /
• pp.337-345
• /
• 2015

현대 사회는 기술의 고도화와 도시화로 인하여 재난의 발생빈도가 높아짐에 따라 재난대비를 위한 재난교육 및 훈련이 중요성이 강조되고 있다. 본 연구에서는 모의훈련과 의식조사를 통한 안전관리 방안을 제시하고자 한다. 이에 본 연구를 통하여 얻어진 개선방안은 첫째 본인의 역할과 책임부여을 위한 교육과 훈련을 실시하여야 한다. 둘째 재난관리와 위기관리의 확대를 위해서 매뉴얼의 구축과 이를 기반으로 한 모의훈련이 필요하다. 셋째 재난관리가 우선시 되지 못하는 조직의 대하여 인력배치 등 재구성을 검토하여야 한다.

• 수산해양기술연구
• /
• 제51권4호
• /
• pp.504-511
• /
• 2015

In this paper, the results of evaluating the passenger comfort due to the standard deviation of acceleration in vertical and lateral direction regarding the ship response in irregular wave by ordinary strip method in regular wave and energy spectrum using linear superposition theory in order to evaluate the motion of experimental ship are as follows. According to the results of ship response, it was possible to find that, in order to reduce the motion of ship, a ship operating in bow sea was more stable than in quartering sea. In the results of analyzing the standard deviation of acceleration in vertical direction according to each component wave pattern, when there was a wave length of 56m and an average wave period of 6 sec, most of cases showed the peak value. And among them, the standard deviation was 0.35 which was the highest in head sea. And in case of lateral direction, the maximum value was shown in a wave length of 100m and an average wave period of 8 sec. And it was 0.16 in beam sea and ${\chi}=150^{\circ}$. In the evaluation of passenger comfort due to standard acceleration in vertical and lateral direction, it was 80% in head and bow sea. On the other hand, it was shown to be 15% in follow sea. Accordingly, when the expected wave height in a sea area where a training ship was intended to operate was known, it was possible to predict the routing of ship. And altering her course could reduce the passenger comfort by approximately 50%.

• 한국정보통신학회:학술대회논문집
• /
• 한국해양정보통신학회 2008년도 춘계종합학술대회 A
• /
• pp.229-232
• /
• 2008

A Neural network classification of human activity data is presented. The data acquisition system involves a tri-axial accelerometer in wireless sensor network environment. The wireless ad-hoc system has the advantage of small size, convenience for wearability and cost effectiveness. The system can further improve the range of user mobility with the inclusion of ad-hoc environment. The classification is based on the frequencies of the involved activities. The most significant Fast Fourier coefficients, of the acceleration of the body movement, are used for classification of the daily activities like, Rest walk and Run. A supervised learning approach is used. The work presents classification accuracy with the available fast batch training algorithms i.e. Levenberg-Marquardt and Resilient back propagation scheme is used for training and calculation of accuracy.

• 전자통신동향분석
• /
• 제33권4호
• /
• pp.1-13
• /
• 2018

Deep learning (DL) models have been widely applied to AI applications such image recognition and language translation with big data. Recently, DL models have becomes larger and more complicated, and have merged together. For the accelerated training of a large-scale deep learning model, model parallelism that partitions the model parameters for non-shared parallel access and updates across multiple machines was provided by a few distributed deep learning frameworks. Model parallelism as a training acceleration method, however, is not as commonly used as data parallelism owing to the difficulty of efficient model parallelism. This paper provides a comprehensive survey of the state of the art in model parallelism by comparing the implementation technologies in several deep learning frameworks that support model parallelism, and suggests a future research directions for improving model parallelism technology.

• 한국정보통신학회논문지
• /
• 제25권3호
• /
• pp.490-493
• /
• 2021

This paper proposes an embedded system that detects mask and face recognition based on a microprocessor instead of Nvidia Jetson Board what is popular development kit. We use a class of efficient models called Mobilenets for mobile and embedded vision applications. MobileNets are based on a streamlined architechture that uses depthwise separable convolutions to build light weight deep neural networks. The device used a Maix development board with CNN hardware acceleration function, and the training model used MobileNet_V2 based SSD(Single Shot Multibox Detector) optimized for mobile devices. To make training model, 7553 face data from Kaggle are used. As a result of test dataset, the AUC (Area Under The Curve) value is as high as 0.98.

• 한국항공운항학회지
• /
• 제30권3호
• /
• pp.109-116
• /
• 2022

A modular platform development technique was proposed to minimize development cost and development period by utilizing the already developed unmanned Aerial target AVT, which has been operated and verified for many years. New Mission Profile was designed and structural analysis was performed through finite element analysis (FEA) by analyzing mission requirements for visual short-range, non-visible mid-range, and long-range targets. The targets are used for guided missile anti-aircraft training. In addition, avionics systems including flight control computers for autonomous flights were developed to verify their conformance by performing launcher take-off tests with rapid acceleration changes and autonomous flight tests at a maximum speed of 300km per hour.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2021년도 추계학술발표대회
• /
• pp.446-448
• /
• 2021

In training a deep learning model, it is crucial to tune various hyperparameters and gain speed and accuracy. While hyperparameters that mathematically induce convergence impact training speed, system parameters that affect host-to-device transfer are also crucial. Therefore, it is important to properly tune and select parameters that influence the data loader as a system parameter in overall time acceleration. We propose an automated framework called Num Worker Tuner (NWT) to address this problem. This method finds the appropriate number of multi-processing subprocesses through the search space and accelerates the learning through the number of subprocesses. Furthermore, this method allows memory efficiency and speed-up by tuning the system-dependent parameter, the number of multi-process spawns.

• 재활복지공학회논문지
• /
• 제8권4호
• /
• pp.275-281
• /
• 2014

본 연구에서는 가이드 폭과 위치조절을 통해 퍼팅 시 순차적인 집중력 강화를 유도하여 퍼팅능력향상을 이끌어낼 수 있는 퍼팅훈련시스템을 개발하고 유용성을 평가하였다. 시스템의 하드웨어는 LEGO NXT 콘트롤러와 금속 Tetrix 세트를 이용하여 개발하였으며 소프트웨어는 LabVIEW ver. 2010을 사용하였다. 시스템의 유용성 평가를 위해 비숙련자 10인과 숙련자(KPGA) 5인에 대해 개발된 시스템의 폭 위치 변화 퍼팅훈련을 10세트 실시하고 의견을 조사하였다. 실험결과 손목부위에서의 속도 가속도 변화 감소 및 머리부위의 변위 감소를 확인할 수 있었으며, 의견조사에서 집중력 향상과 동기부여에 도움이 된다는 소견을 얻을 수 있었다. 본 시스템의 가이드 폭 위치 조절방식의 훈련법은 퍼팅을 위한 것으로 비슷한 운동구조를 갖는 노약자 장애인용 파크골프나 게이트볼 훈련에도 쉽게 적용이 가능할 것으로 판단된다.

• 한국항행학회논문지
• /
• 제27권2호
• /
• pp.155-161
• /
• 2023

비행착각(Vertigo)이란 공간상에서 헬리콥터의 위치, 자세, 움직임 등과 관련된 인지가 부족한 상태를 일컫는다. 짙은 안개 속이나 야간비행 등, 지평선이 보이지 않는 상황에서 비행할 때 비행착각에 빠지기 쉽고 시야가 넓더라도 구름 모양이나 바람 등 기상 조건, 지상물의 상태 등 시각적인 원인, 기체의 자세나 중력가속도의 변화 등과 같은 감각적인 원인에 의해 빠지기도 한다. 조종사 비행 훈련에 있어 지각 및 감각을 요구하는 헬리콥터 비행 훈련의 특수성에 따라 조종사 훈련을 위해 기존의 상용 비행 시뮬레이터 프로그램에 6축 모션 시스템을 적용한 모의비행훈련장치 모션 시스템의 설계와 프로그램에 관하여 연구하였다. 본 연구를 통해 제작된 모션 기반 헬리콥터 시뮬레이터를 활용하여 조종사의 훈련에 활용할 경우 기존에 활용되던 시각 기반 모의비행훈련장치에서 높은 성과를 확인할 수 없었던 비행착각 예방에 긍정적인 효과가 나올 것으로 예상된다.

• 센서학회지
• /
• 제27권2호
• /
• pp.86-92
• /
• 2018

In recent times, motion capture technology using inertial measurement unit (IMU) sensors has been actively used in sports. In this study, we developed a canoe paddle, installed with an IMU and a water level sensor, as a system tool for training and calibration purposes in water sports. The hardware was fabricated to control an attitude heading reference system (AHRS) module, a water level sensor, a communication module, and a wireless charging circuit. We also developed an application program for the mobile device that processes paddling motion data from the paddling operation and also visualizes it. An AHRS module with acceleration, gyro, and geomagnetic sensors each having three axes, and a resistive water level sensor that senses the immersion depth in the water of the paddle represented the paddle motion. The motion data transmitted from the paddle device is internally decoded and classified by the application program in the mobile device to perform visualization and to operate functions of the mobile training/correction system. To conclude, we tried to provide mobile knowledge service through paddle sport data using this technique. The developed system works reasonably well to be used as a basic training and posture correction tool for paddle sports; the transmission delay time of the sensor system is measured within 90 ms, and it shows that there is no complication in its practical usage.